banner
Nagi-ovo

Nagi-ovo

Breezing homepage: nagi.fun
github
主页ArchivesPortfolios日寄友链

多模态

cover
cover
cover
cover
cover
cover
cover
cover
cover
cover
cover
cover
cover
cover
cover

以 Paligemma 为例的视觉语言模型

视觉语言模型可以分为四类^{[1]}： 将图像转为可以和文本 token 共同训练的嵌入特征，如 VisualBERT、SimVLM、CM3 and etc. 学习良好的图像嵌入，作为冻结的预训练预言模型的前缀，如 ClipCap 通过专门涉及的 cross-attention…
深度学习
Ownership of this blog data is guaranteed by blockchain and smart contracts to the creator alone.